Problem with IR2113 full-bridge

Hello guys,

I am working on a MOSFET full bridge with IR2113. The thing seems to be working but with a problem. MOSFETs have a really long turn-off delay (see the oscilloscope attachment). At low frequencies this is not a problem, however, when I increase the frequency to a few kHz the turn-off delay becomes problematic.

Can anyone please help me with the problem. I have attached the circuit diagram and oscilloscope images.

Thanks,

Greg.

At 4 kHz:
View attachment H_Bridge_4kHz.pdf

At 19 kHz:
View attachment H_Bridge_19kHz.pdf

Circuit diagram:
View attachment H-Bridge_2xIR2113.pdf

First R1 and R5 should not be pulled up. They should be across the gate and source as they are on the bottom fets. They're likely to destroy something or be destroyed in that position with 300V across them.

But if green is your switch node (you don't say) something is wrong or busted. The low side fets appear to never turning on. Check your gate drive signals into and out of the driver (is that yellow?).

Hi,

Besides the datasheets I recommend od read application notes.
Especially when you encounter problems with your circuit.

--> AN-978
Especially Figure 28A

Klaus

Hi,

yes, the circuit was not correctly drawn. In the real circuit, both of resistors are connected as in schematic. Probing is done on gates of HI and LO MOSFETs. The problem seems to be on the HI side, not the low side.

Corrected schematics:
View attachment H-Bridge_2xIR2113.pdf

Thank you for all the answers.

you are not measuring, or seeing, the gate drive correctly on your scope - the Vds waveforms show fast turn off - ...

also the waveforms are out of synch - so you are really wrong there ....

Probing is done on gates of HI and LO MOSFETs.

Click to expand...

How connected? What kind of probes? High side gate drive measurement needs a differential probe, otherwise you are measuring Vgs,HS + Vds,LS. The oscilloscope screenshot shows 6V magnitude, how's that possible with 12V driver supply? What's the bridge supply voltage?

Hello I also use this chip, a couple of comments on your schematic
1/ You show no decoupling on VCC and VSS pins, they are essential
2/ You do not mention the diode type, the boost diode is critical, I have not found it necessary nor desirable to use fast turnoff diodes but that is application (and mosfet) dependent.
3/ I have not found it necessary to use gate pulldowns and in particular these load the boost supply
4/ I just noticed your H-bridge power rail is the same as VCC OUCH!! this needs attending to or your load switching transients will upset the IRS2113's, you need to properly divide power paths and keep a separate decoupled reasonably quiet VCC for the driver chips.

Going back to (1) The VCC bulk decoupler must provide the peak charge current for the boost capacitor, I use a small resistor in series with the boost diode to reduce the peak current and hence VCC glitch at low side turn-on.

Lastly layout is critical but I expect you know that

@asdf44 & @KlausST: I was posting wrong (OLD) schematic. Actual circuit has that integrated into design.
@KlausST: According to your answer in this thread (https://www.edaboard.com/showthread.php?362476-IR2113-Mosfet-driver&highlight=ir2113):
Re: IR2113 Mosfet driver
Hi, don´t measure HO wrt GND --> measure HO wrt VS. Is it still equal?
-----------------------------------------------------------------------
I tried that. The signal is basically the same.
Question: So was I basically shorting Vs and GND through the oscilloscope?

@Easy peasy: the input waveforms from function generator are 180 degrees phase shifted squarewaves and are fed on IR2113. On gates the time period of the signal on the upper MOSFETs is somewhat changed. @4kHz there is this slope (of which I am mainly concerned about) but time period remains the same and @19kHz slope remains and time period lenghtens.

@Fvm: I use Hantek 6022BE with x10 probes and mistake comes from not selecting right probe settings in software and picture printout which does not show grid. Since grid of osciloscope isn't displayed, you cannot see divisions. 5xtick=one division -> 6xtick -> 12V @ 12V/div. Should I use AC coupling oscilloscope? Or is it quite the same result if I measure from VS to HO? At least based on my understanding it would be the same.

@fourtytwo:
1) Are they a must be? And where(Power supply, close to IR2113, anywhere else? Decoupling capacitors are mainly implemented to compensate for stray inductances, right? I'd appreciate good read on decoupling, I googled but didn't get the answers, that would help me give a good estimate of values. For power supplies or voltage regulators 100uF,100nF parallel was often mentioned. Are 100uF && 100nF parallel good selection?

2) All diodes are UF4007. I saw somewhere on forums that this is "slow" diode and shouldn't be used for arround 100kHz. They should definately handle 4kHz operation, don't you think? I was considering bootstrap circuit is causing problems, so I added resistance as AN-6076 suggests, tried various cap values but problems remains the same.
Value of bootstrap capacitor was calculated from AND9674/D(AN-6076) arround 100nF. KlausST suggested AN-978. UF4007 are parallel with gate resistor.

3) These resistors are suggested for achieving higher frequencies(which I am after) and also to minimize switching losses.

4) Also crossed my mind. Due to safety precautions and in lack of 300V DC supply I tried 12V for rail supply. I figured when this will work I will step up. I am currently supplying 5V and 12V from computer power supply and voltage rail from adapter 230VAC/12VDC@2A. Sh

Really appreciate the help, but my mind was also boggled by similar remarks. It could also be function generator circuit. It comes from DDS, which outputs "main" signal and inverted signal, is amplified through summing amplifier and then fed to IR2113. There is basically no "dead time" between LO and HI signal change. Is this a problem, or should
IR2113 handle that?

I have not yet tried different diodes, decoupling caps and didn't yet improve the layout, was hoping to improve layout once I get better results.
I will make new PCB from scratch, perhaps I miss-looked something.

Hi,

Question: So was I basically shorting Vs and GND through the oscilloscope?

Click to expand...

We can not answer this, because we don´t know about your scope, your used probes, your wiring....

Klaus

Btw:
* UF4007 are fast. 1N4007 are slow.